/srv/irclogs.ubuntu.com/2011/12/15/#ubuntu-x.txt

RAOFcnd: Can you keep the git branch called ubuntu+1 for now, and you don't need to use a standard suffix.  Treat the staging PPA as the main archive, just delayed :)00:04
RAOFcnd https://launchpad.net/~ubuntu-x-swat/+archive/x-staging00:05
cndRAOF, ok, I'll be uploading xorg-server_2:1.11.2.902-1ubuntu100:05
RAOFbryceh: Could you add cnd to ~ubuntu-x-swat?00:05
cnddoes that look right?00:05
cndjust double checking :)00:05
RAOFThat looks right to me, yes.00:05
brycehsure00:08
brycehlooks like someone beat me00:10
RAOFCool ;)00:14
cndRAOF, xorg-server has been accepted into ppa:ubuntu-x-swat/x-staging00:36
cndreminder that I completely have not tested the code base :)00:36
cndit's *your* fault if your input devices catch on fire :)00:37
RAOFcnd: Excellent!  I'll start throwing drivers at it.00:37
RAOFHah.  The way my hardware luck's going, *they will*00:37
cndhooray! It's built01:31
RAOFLet the wave of babies begin!01:32
Sarvattcnd: who needed pinging about unity not working again with it because of libutouch-geiss?01:34
RAOFDBO or bregma01:34
cndSarvatt, it's either an issue with the utouch stack, or an issue with unity misusing it01:35
Sarvatthttps://bugs.launchpad.net/ubuntu/+source/unity/+bug/860707 comment #5 will be whats hit with whats in there01:35
ubot4Launchpad bug 860707 in unity (Ubuntu) (and 1 other project) "Unity crashes when started in an environment without utouch support (affects: 58) (dups: 2) (heat: 284)" [High,Confirmed]01:35
Sarvattunity side was fixed01:35
cndthen bregma would be the one01:35
Sarvattby http://bazaar.launchpad.net/~unity-team/unity/trunk/revision/1728#plugins/unityshell/src/GeisAdapter.cpp01:35
cndSarvatt, that unity code doesn't look right to me...01:36
cndwait, nm01:36
cndread the diff wrong01:36
Sarvatti worked around it locally by ripping geis support out of unity so maybe its still a unity bug01:37
Sarvatt"ripping geis support out" == commenting out 2 lines..01:38
cndSarvatt, it looks like unity is still trying to use geis01:38
cndeven when it's not supported01:38
cndotherwise it shouldn't be calling geis_configuration_get_value01:38
cndso DBO or MacSlow (Mirco, who wrote that commit) should be pinged01:38
* cnd is opening the code to peek01:39
Sarvatthttp://sarvatt.com/downloads/patches/byegeis.patch was what was needed01:40
cndSarvatt, GeisAdapter::_root_instance is never initialized01:41
cndit must be initialized to nullptr01:41
cndSarvatt, can you follow up on it?01:41
cndI'm on vacation now through  monday01:41
Sarvattshould be easier with this ppa, ppa-purge is busted with multiarch so it took me a few hours to switch fixing it up when i was doing it in edgers01:42
cndSarvatt, I would suggest hitting MacSlow with a clue-by-four to get it resolved :)01:43
cndyou can leave bregma and DBO unharmed01:43
Sarvattcnd: yeah sorry to bug ya on vacation man, was just picking your brain trying to find out who to bug :)01:43
cndwell, technically I am still working for 15 mins :)01:44
cndI just can't follow up on it01:44
* Sarvatt is on vacation now too, i know how it goes :)01:44
cndahh, ok01:44
cndI can send an email right now then01:44
cndyeah, I'll just do that01:44
cndSarvatt, I updated the bug with all the info, hopefully that's enough to see it through01:51
Sarvattya asked me before and i didn't follow through because reproducing it with edgers is really a pain in the ass, i'm really sorry about that man01:52
Sarvattthanks :)01:52
cndSarvatt, np :)01:55
cndif I had looked closely enough it would have been obvious01:55
cndshould have done it sooner01:55
Sarvattedgers was broken for 2 months because i didn't comment out 2 lines of code, i know the feeling01:56
Sarvattso let me get this straight, we're never going to be able to update xserver at a decent schedule anymore because fglrx taking 3 months to adapt to new abis and breaking unity with fglrx isn't an option because DX uses them?02:29
Sarvattonly the xserver releases where xserver releases 3 months before 2 months before final freeze02:30
RAOFSarvatt: No, this is a one-time (or at least LTS-only) deal; it was quite clear in the sessions at UDS that “use an open-source driver” is an acceptable work around.02:45
Sarvattthe +1 will work at all times effort doesn't seem LTS only though02:45
Sarvattoh ok02:45
bjsnideris the radeon driver a good alternative to fglrx at this point?02:46
Sarvattvery much so02:49
Sarvattespecially for the desktop where it works much better than fglrx02:49
Sarvattxv is busted on fglrx in the december release, people using it get to wait a month for a fix02:50
bjsniderdoesn't say much for amd02:51
Sarvattagreed :)02:51
bjsniderit works "much better"?02:52
bjsniderthat's mind-boggling02:52
SarvattMUCH02:52
Sarvattso much better it isnt funny, they keep implementing things like the tearing fix which is broken and it takes them a month to fix what they break02:53
bjsniderdoes opengl work much better in radeon?02:53
Sarvattthey work on 3 monthly branches at a time, fixes go in the newest one, if its really important it might make release+1 version but usually not02:54
Sarvattand for the past year its been working very bad imo02:54
Sarvattthere are no stable updates so you are guaranteed to have to wait a month for a fix if somethings broken02:56
Sarvatt(at least)02:56
Sarvattmore likely 2, worst case 302:56
Sarvattbjsnider: for games sure02:58
Sarvattthats what i care about so i'd have to put up with it02:58
Sarvattbut if you just care about the desktop the open source ones are a much better option02:58
bjsniderfor games fglrx is the better option?02:58
Sarvattsane as nvidia, yeah much02:58
Sarvatts/sane/same/02:59
bjsniderwell, i guess that's what the firegl customers care about02:59
bjsnideropengl02:59
Sarvattfiregl customers care about professional 3D apps02:59
bjsniderand that's the only thing fglrx does well i guess03:00
Sarvattand yea thats much faster, thats what fglrx was even released for, not much to do with a windows game in wine :P03:00
bjsniderwhat about mobile chips?03:01
RAOFIf we can enable float textures & add libdxtn, radeon & intel should be reasonable for wine games, too.03:01
Sarvattbjsnider: what about mobile chips? they're usually just a speed bump behind desktop ones03:02
RAOFDunno about nouveau, but probably ok too.03:02
Sarvattnouveau scares me03:02
bjsnideri meant that you qualified your remarks by saying that radeon is much better on desktops, as if mobile stuff was different03:03
Sarvattits broken on lots of crap, especially in mesa03:03
Sarvattbjsnider: nope mobile/desktop doesnt make a difference, its better for unity on everything03:03
RAOFIt *is* much harder to recommend nouveau over nvidia than it is to recommend radeon over fglrx :)03:04
SarvattRAOF: we hit so many problems and have noone to escallate bugs to03:04
SarvattQA runs tests against nouveau, suspend/resume is completely busted03:04
brycehactually we did meet the nouveau devs at the X conference, and can escalate to them03:04
brycehit's just that it's fairly hard for them to solve bugs03:05
bjsniderthere's a public bug tracking system for nouveau isn't there?03:05
brycehyes03:05
Sarvattnouveau devs = darktama though, escallating bugs to another distro for unreleased hardware?03:05
Sarvattfor mesa when he works on the kernel mostly?03:05
bjsniderwhy is it fairly hard for them to solve bugs?03:06
brycehbjsnider, lack of access to register documentation03:07
brycehbjsnider, and lack of man power for reverse engineering03:07
bjsniderthere hasn't been any amd documentation in 3 years i think03:07
brycehbjsnider, true but AMD employs several of the -ati developers who (AIUI) can get some insider info as a result03:07
Sarvattnouveau to me is really done on a best effort basis, we really rely on the vendors a lot to fix bugs on unreleased stuff which i work on03:07
bjsniderthat doesn't inspire a lot of confidence in nouveau03:10
bjsniderwhat happens when everybody and their dog wants to switch to wayland? you need nouveau for that03:10
RAOFEh.  Suspend/resume is broken with nvidia, too :)03:13
SarvattRAOF: there's always a corner case, for the most part it shouldn't be though03:13
Sarvattaka for crap thats certified its definitely not03:14
RAOFWell, my experience with nouveau has tended to include working suspend/resume.  Even on this t420s (before it stopped POSTing ☹)03:14
Sarvattwait, nouveau?03:14
Sarvattthats specifically the platform i'm saying suspend/resume is broken on nouveau on :)03:14
RAOFSarvatt: It was working for me at one point, then it stopped working.  But it stopped working on nvidia at the same time :)03:15
Sarvatthmm03:16
Sarvattprecise?03:16
Sarvattjust curious because thats gonna be a huge problem if it is03:16
Sarvattso those systems shipped with 10.10, with acpi_osi="!Windows 2009" so it pretended to be XP so optimus wouldn't be used03:17
RAOFI installed that image, and that's what it came up with.  Then I suspended, and it didn't feel like waking.  Then I reset the BIOS options, and it didn't feel like POSTing :/03:18
Sarvattinstalled what image?03:20
Sarvatti have a freaking weird job, i need to identify workarounds need to make specific systems work so they can be preinstalled, the make them work in that release+1 out of the box03:21
bjsniderdoes suspend/resume work well using the blob?03:44
bjsnideri read complaints03:44
RAOFbroder: Confirmed; hybrid-detect only prints "integrated" on the intel/ati hybrid system, even with the break removed (so, iterating over all the PCI devices)04:54
broderRAOF: excellent, thanks05:41
broderwhat happens if i start X with both nvidia-current and nvidia-173 installed?07:30
broderhow does it figure out which kernel module to use?07:30
tjaaltondpkg-checkbuilddeps: error: --help is not a valid version'07:31
tjaaltonfun07:31
tjaaltonBuild-Depends: debhelper (>= 8), dh-autoreconf, pkg-config, xserver-xorg-dev (>= --help)07:33
tjaalton:P07:33
ricotzcnd, hi, i think you should bump this depend x11proto-input-dev (>= 2.0.1-1ubuntu1) in 2:1.11.2.902-1ubuntu107:36
Sarvattricotz: configure.ac require higher?07:38
ricotzRAOF, hi, ^07:38
ricotzSarvatt, i guess it is mandatory for the backport xi2.2 stuff07:39
ricotzso i guess the configure bump is missing too07:39
Sarvattis it mandtory? a commit got left out of the backport if so07:40
ricotzi think so07:41
tjaaltonpatch 05/42 of the multitouch bundle07:42
=== jibel_ is now known as jibel
=== yofel_ is now known as yofel
ricotztjaalton, hi, do you have a link to your libwacom repo if you already pushed it?14:07
tjaaltonricotz: haven't pushed yet14:16
ricotztjaalton, ok14:17
tjaaltonhum, left already14:30
tjaaltongit://anonscm.debian.org/users/tjaalton-guest/libwacom.git14:30
RAOFtjaalton: Yeah, I spotted that *almost* immediately ;)21:07
RAOFA suprising amount of our desktop really, really doesn't like multitouch going away.22:01
RAOFUnity?  Dies.  Unity2D?  The launcher will spin at 100% CPU and not display.  Evince?  Likewise.22:02
FernandoMigueleheh22:05
jcristauso you can't remote an ubuntu desktop against a $something_else X server?22:06
RAOFI guess not.22:10
RAOFThose are all bugs, but they're obviously bugs that no-one's cared enough about; I think it's a pretty rare corner-case.22:11
SarvattRAOF: the evince one popped up in a precise libutouch-geis update, ricotz has a bug open about it22:14
Sarvattthink it affected EOG too22:14
Sarvatthttps://bugs.launchpad.net/ubuntu/+source/utouch-geis/+bug/89817522:20
ubot4Launchpad bug 898175 in utouch-geis (Ubuntu) "uTouch hangs using XCB back end if X server does not support gesture protocol (affects: 3) (heat: 16)" [Undecided,Confirmed]22:20
Sarvatthttps://bugs.launchpad.net/ubuntu/+source/utouch-geis/+bug/87934822:20
ubot4Launchpad bug 879348 in utouch-geis (Ubuntu) (and 1 other project) "evince segfaults (affects: 8) (dups: 2) (heat: 49)" [Undecided,Incomplete]22:20
SarvattRAOF: finally friggin found it http://www.spinics.net/lists/dri-devel/msg17272.html22:59
Sarvattit was on dri-devel, i was looking through lkml and intel-gfx22:59
RAOFBecause having just a single mailing list would have made that too hard ;)23:02
Sarvattoh boy, file system corruption for that guy from it too23:02
RAOFYeah.  Memory corruption is not my favourite thing.23:03
RAOFAs you can see, I'm not up-to-date on my dri-devel reading :)23:05

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!